Caleb Kruse

Create YouTube thumbnail concepts with Nano Banana 2 using image references for a person, product, or brand. The workflow applies prompt formulas for recognizable layouts and likeness, generates multiple variations in parallel, and helps review them for readability and visual impact.

中文简介正在准备中。当前显示英语简介。

• Create a thumbnail for a YouTube video
• Generate variations for a thumbnail A/B test
• Use face and brand reference images in a thumbnail

Analyze a reference video’s structure, pacing, camera work, editing, and tone, then turn its reusable style patterns into a Seedance 2.0 prompt template with variables and examples. The template is designed around the model’s 15-second clip limit.

中文简介正在准备中。当前显示英语简介。

• Turn a reference video’s style into a reusable prompt template
• Identify the key beats and camera choices in a video
• Adapt a longer video style to short generated clips

Turn an existing image ad into a reusable prompt template with replaceable brand details. Analyze its layout and text, generate and refine a match, then test the generalized template with another brand and record model behavior in the shared library.

中文简介正在准备中。当前显示英语简介。

• Turn a reference ad into a reusable prompt
• Adapt an ad layout for another brand
• Compare template results across image models

Generate or edit standalone Meta ad images through Arcads with ChatGPT Image 2. This skill is suited to text-heavy layouts and UI-style concepts such as chat threads, comparison tables, and dialog screens, then provides files for review and handoff.

中文简介正在准备中。当前显示英语简介。

• Creating an ad that resembles a chat conversation
• Making a comparison table or search-results image ad
• Editing text-heavy static ad creatives

Use a finished image or video to prepare a Meta ad, write multiple copy variations, and place it in an existing ad set as a paused ad. The workflow can also review top-performing account ads by ROAS or research competitor ads in the Ad Library.

中文简介正在准备中。当前显示英语简介。

• Prepare a finished image for a Meta ad
• Create ad copy and add a video to an existing ad set
• Review top ads by ROAS before writing new copy

Analyze a source video ad’s beats, visuals, dialogue, and tone, then adapt those patterns for a different product or offer and generate a Seedance 2.0 video. Longer ads can be split into short clips, with dialogue and product details tailored to the new brief.

中文简介正在准备中。当前显示英语简介。

• Adapt a video ad’s format for a different product
• Recreate a source ad’s pacing and camera style
• Generate a short video ad from a reference clip and product image

Use this skill to create and retrieve video, image, scene, and script-related assets through the Arcads external API. It guides model and prompt selection, API authentication, and polling generated assets until they are ready.

中文简介正在准备中。当前显示英语简介。

• Generating a product video with an AI model
• Creating still images or B-roll through Arcads
• Retrieving generated assets and checking their status

Create or edit standalone Meta ad images through Arcads using Nano Banana models. This skill helps shape prompts for photoreal and lifestyle concepts, applies layout safeguards, and returns generated image files for review and handoff.

中文简介正在准备中。当前显示英语简介。

• Making a lifestyle image ad with Nano Banana
• Editing an existing ad image or background
• Creating product creatives from reference images